Sounds of valves in heating systems for classification and condition monitoring

Published: 20 September 2022| Version 2 | DOI: 10.17632/y6fkrybb32.2
Contributors:
,
,

Description

Dataset contains 427 sounds of valves installed and operating in district heating systems. Tables in Excel and Matlab format contain categories for each sound as follows: 1) FlowNoise, 2) Rattling, 3) Whistling, 4) Cavitation. 'FlowNoise' class corresponds to normal valve operation. The other three classes correspond to unwanted operating conditions. Attached Matlab table contains also 65 extracted features (acoustic features and statistical features) for each sound. The dataset was prepared to train machine learning classifiers to classify sounds into corresponding categories. FILE DESCRIPTION: 1) 'extracted_features_and_classes.mat' Matlab 2019 format (table), includes: - mp3 file names - corresponding categories (FlowNoise, Rattling, Whistling, Cavitation) - 65 extracted features for each mp3 file 2) 'mp3_file_list_and_classes.xlsx' Excel format, includes: - mp3 file names - corresponding categories (FlowNoise, Rattling, Whistling, Cavitation) - links to mp3 files to play sounds (mp3 files should be unzipped in the same folder) 3) 'mp3_files.zip' ZIP archive with 427 sound files in mp3 format. Each sound has duration of 3 seconds.

Files

Institutions

Univerza v Ljubljani Fakulteta za strojnistvo

Categories

Machine Learning, Feature Extraction, Control Valve, Vibration Condition Monitoring

Licence